why 6667 usec is divided by 512 and not 8? from what i recall you don't want too low duty cycle or display will be too dim even with huge LED current.
1/8 is fine, doing 1/16 is pushing it, i didn't try it but i would expect that 1/512 is not good idea.
the process is something like:
1. turn on one plane (one of 8 cathodes)
2. load 64 bits (anodes)
3. wait short time
4. increment plane number
5. repeat
how you get 64 outputs for anodes is up to you (octal latches, SIPO registers etc...)
to control brightness of entire display, you could connect MCU PWM output channel to enable input on TTL chips.
to control brightness of individual LED in a plane, you need to control it in software
here is an example using multiplexer to light up one plane at a time (64 LEDs):
https://www.instructables.com/id/Led-Cube-8x8x8/
and here is the part of circuit to load 64 bits:
https://www.instructables.com/id/Led-Cube-8x8x8/step8/IO-port-expansion-more-multiplexing/
they used parallel loading of 8-bit and in only few clocks, all 64 anodes are loaded with corresponding bit-pattern.
cathodes are cycled at 1/8 duty cycle (one plane at a time)